You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@cassandra.apache.org by "Antonio (Jira)" <ji...@apache.org> on 2020/02/24 03:47:00 UTC
[jira] [Created] (CASSANDRA-15593) seems reading repair bug
Antonio created CASSANDRA-15593:
-----------------------------------
Summary: seems reading repair bug
Key: CASSANDRA-15593
URL: https://issues.apache.org/jira/browse/CASSANDRA-15593
Project: Cassandra
Issue Type: Bug
Reporter: Antonio
Assignee: Aleksey Yeschenko
cassandra version:2.1.15
i have one dc and 3nodes
1. create KEYSPACE test WITH replication = \{'class': 'NetworkTopologyStrategy', 'DC1':'3' } and durable_writes = 'true';
2. create table test(a int , b int , c int , primary key(a)) with dclocal_read_repair_chance = 1.0;
3. insert one row into table test,instert into test(a, b, c) values (1, 1, 1); and remove sstable on two nodes and result below:
node1:have correct row 1 1 1
node2:doesn't have rf
node3:doesn't have rf
4. and i use local_one select one by one like this:
node1 un,node2 dn,node3 dn:return 1 1 1
node1 dn,node2 un,node3 dn:return null
node1 dn,node2 dn,node3 dn:return null
this prove node1 have correct rf
5. and i let all node un,user local_quarum to select , select * from test where a = 1;
but the read repair does't work every time(i test many times),that's the problem(same in 3.0.15)
i hope if i set dclocal_read_repair_chance = 1.0,every time i read by local_quarum, if any rf digest does't match,read repair will work,and repair all nodes
i.m not sure does's the problem happends in this code()
wish for your reply,thanks
public void response(MessageIn<TMessage> message)
{
resolver.preprocess(message); int n = waitingFor(message)
? recievedUpdater.incrementAndGet(this)
: received; if (n >= blockfor && resolver.isDataPresent())
{
////this mean if return responses >= rf/2 +1 and a data response return,it start compare,does't all response
condition.signalAll(); // kick off a background digest comparison if this is a result that (may have) arrived after // the original resolve that get() kicks off as soon as the condition is signaled if (blockfor < endpoints.size() && n == endpoints.size())
{
TraceState traceState = Tracing.instance.get(); if (traceState != null)
traceState.trace("Initiating read-repair");
StageManager.getStage(Stage.READ_REPAIR).execute(new AsyncRepairRunner(traceState));
}
}
}
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@cassandra.apache.org
For additional commands, e-mail: commits-help@cassandra.apache.org